Skip to main content
NCBI home page
EPA Author Manuscripts logoLink to EPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Dec 17.
Published in final edited form as: J Ecohydraul. 2020 Dec 17;1:1–13. doi: 10.1080/24705357.2020.1855086

Assessing contributions of cold-water refuges to reproductive migration corridor conditions for adult Chinook Salmon and steelhead trout in the Columbia River, USA

Marcía N Snyder 1, Nathan H Schumaker 1, Jason B Dunham 2, Matthew L Keefer 3, Peter Leinenbach 4, Allen Brookes 1, John Palmer 4, Jennifer Wu 4, Druscilla Keenan 4, Joseph L Ebersole 1
PMCID: PMC8059528  NIHMSID: NIHMS1690098  PMID: 33898904

Abstract

Diadromous fish populations face multiple challenges along their migratory routes. These challenges include suboptimal water quality, harvest, and barriers to longitudinal and lateral connectivity. Interactions among factors influencing migration success make it challenging to assess management options for improving migratory fish conditions along riverine migration corridors. We describe a spatially explicit simulation model that integrates complex individual behaviors of fall-run Chinook Salmon (Oncorhynchus tshawytscha) and summer-run steelhead trout (O. mykiss) during migration, responds to variable habitat conditions over a large extent of the Columbia River, and links migration corridor conditions to fish condition outcomes. The model is built around a mechanistic behavioral decision tree that drives individual interactions of fish within their simulated environments. By simulating several thermalscapes with alternative scenarios of thermal refuge availability, we examined how behavioral thermoregulation in cold-water refuges influenced migrating fish conditions. Outcomes of the migration corridor simulation model show that cold-water refuges can provide relief from exposure to high water temperatures, but do not substantially contribute to energy conservation by migrating adults. Simulated cooling of the Columbia River decreased reliance on cold-water refuges and there were slight reductions in migratory energy expenditure. This modeling of simulated thermalscapes provides a framework for assessing the contribution of cold-water refuges to the success of migrating fishes, but any final determination will depend on analyzing fish survival and health for their entire migration, water temperature management goals and species recovery targets.

Keywords: water temperature, migration, behavioral thermoregulation, individual-based model, refuge

Introduction

Migration is a strategy that freshwater fishes use to exploit diverse reproductive, rearing, or feeding habitats, and in intact systems, can result in benefits including enhanced body size, reproductive output, and population size (Gross 1988). However, migration can incur substantial costs. For diadromous fishes that use large river corridors during migration, costs can include the energetic costs of navigating barriers, avoiding predators, and minimizing exposure to other stressors (Bernatchez and Dodson 1987; Fenkes et al. 2016; Northcote 1984). Thus, the viability of diadromous life histories is in part dependent upon the continued suitability of migratory corridors (McClure et al. 2003).

As rivers warm globally (e.g. Kaushal et al. 2010; Webb and Nobilis 2007), migratory fishes can encounter increased physiological stresses associated with higher water temperatures during migration. For cold-water fishes such as salmon (Salmo and Oncorhynchus spp.) that must migrate through corridors approaching the upper thresholds of thermal tolerance, cold-water refuge (CWR) availability has been shown to be important (Frechette et al. 2018; Keefer et al. 2009). Benefits of behavioral thermoregulation in CWRs include temporary relief from acute temperatures that would otherwise lead to elevated metabolic demands with adverse consequences for gamete development (e.g. Geist et al. 2006), disease resistance (e.g. Crossin et al. 2008), and survival rates (e.g. Bowerman et al. 2017b). Conservation of somatic energy reserves, necessary for lengthy migrations and eventual costs of reproduction, is hypothesized to be an additional benefit of reduced thermal exposure from behavioral thermoregulation (Connor et al. 2019). The value of these cold-water features is expected to become more important in the future as warming trends continue (Crozier et al. 2019; Fullerton et al. 2017; Keefer et al. 2018).

Recognition of the potential importance of CWRs for the viability of migratory life histories has led to calls for protecting and restoring thermal diversity in migratory corridors (Shaftel et al. 2020; Torgersen et al. 2012). In 2003, the US Environmental Protection Agency provided guidance to States and Tribes in the Pacific Northwest USA for setting temperature standards protective of fish ​that included a recommendation to include CWR protections (US Environmental Protection Agency 2003). This was followed by calls for more complete incorporation of the spatial and temporal thermal regime into numeric and narrative temperature criteria (Poole et al. 2004). Incorporating CWRs into water temperature standards, however, has been challenged by the inability to quantify the relative benefits of CWR; specifically, how much CWR habitat is “enough”, and how should CWRs be spaced to allow successful migration and avoidance of thermally stressful conditions? For example, the State of Oregon adopted a water temperature standard that became the first in the USA to incorporate CWRs in its narrative standard alongside the numeric standard for selected migratory corridors (Box 1, OAR 340-041-0028). Application of the narrative standard was hindered, however, in part by the challenge of defining the quantity and distribution of CWRs required to be “sufficient”, and in 2015, the US National Oceanic and Atmospheric Administration issued a jeopardy opinion stating that Oregon’s migration numeric criterion needed application of the supplementary CWR narrative standard in order to protect migrating Pacific salmon and steelhead trout (Oncorhynchus spp.).

Box 1. Policy background.

The designated beneficial use in Oregon’s water quality standards for the Columbia River is “salmon and steelhead migration corridor.” Oregon designated the migration corridor use for large lower mainstem river reaches where the primary and most sensitive aquatic life use is for adult salmon and steelhead migration during the summer, and where lesser or no use for salmonid rearing occurs in the months of July and August. In addition, it is unlikely that the natural thermal conditions of these large lower mainstem reaches would ever have maintained water temperatures below 18°C throughout the summer, with the exception of cold-water refuges (CWR) formed at cold tributary junctions or other sources of cold water such as groundwater or hyporheic exchange. Under Oregon’s plan, the temperature criteria that apply to migration corridors are a 7-day average of maximum temperature at 20°C (68.0 degrees Fahrenheit) and the narrative criterion for CWR. The narrative criterion states that these water bodies must have sufficiently distributed CWR that allow salmon and steelhead migration without significant adverse effects from a rolling weekly average of daily maximum water temperatures up to 20°C.

The narrative criterion calls for CWRs that are sufficiently distributed so that salmon and steelhead migration can occur without significant adverse effects from warmer water temperatures in the main stem Willamette River.

In 2015 the National Marine Fisheries Service (NMFS) released its Endangered Species Act Biological Opinion on the Environmental Protection Agency’s Proposed Approval of Certain Oregon Water Quality Standards Including Temperature and Intergravel Dissolved Oxygen. In the 2015 Biological Opinion, NMFS found jeopardy on the U.S. EPA’s approval of Oregon’s 2003 water temperature criterion for migration corridors. NMFS questioned the protectiveness of the 20°C criterion relative to the Salmon and Trout Rearing and Migration criterion of 18°C, because Oregon’s Department of Environmental Quality (DEQ) had not demonstrated how to interpret the CWR narrative provision. Studies (including this one) are underway to inform the basis for interpreting the narrative criterion and evaluating whether there is sufficient cold-water refuge (CWR), for subsequent application to migration corridors.

To address this need, we developed a simulation model to quantify the relative costs and benefits of CWR use by adult, migratory salmon (Snyder et al. 2019). The current paper extends those authors’ work by using their spatially explicit, individual-based model (IBM) to assess questions of CWR “sufficiency” within the Columbia River corridor in Oregon and Washington, USA. We use the models to evaluate multiple scenarios of CWR availability, that hypothesize both additions and losses along the migratory corridor. Further, we compare the results from these experiments to those obtained when Columbia River corridor temperatures are reduced. Outcomes are assessed in terms of fish energy use and accumulated exposure to acutely warm water temperatures.

Methods

Model overview

To evaluate the relative benefits of behavioral thermoregulation along the Columbia River, we ran simulations using an updated version of the spatially explicit individual-based migration corridor simulation model initially described in Snyder et al. (2019) (Figure 1). The spatial extent of the Snyder et al. (2019) model was expanded from the Bonneville Pool (a 73-rkm-long reservoir) to a 288-km reach of the lower Columbia River from Bonneville Dam to the Snake River confluence (Figure 2). In addition, mechanisms were added to the model to simulate fish passage through fishways at three hydropower dams (The Dalles, John Day, and McNary dams, Figure 2). Furthermore, populations of salmon and steelhead trout from the original model were re-parameterized for the specific salmon and steelhead trout populations of interest.

Figure 1.

Figure 1.

Open in a new tab

Migration corridor simulation model sequence of events modified from Snyder et al. (2019) to include simulation of hydropower passage. Boxes represent model modules. The model processes the sequence of events once per time step (hourly) with the exception of the model initialization which only runs the first time step.

Figure 2.

Figure 2.

Open in a new tab

Map showing A) the Columbia River migration corridor from Bonneville Dam to Snake River confluence with the Snake River Fall Chinook Salmon evolutionary significant unit (ESU) and Grande Ronde River summer steelhead spawning habitat highlighted and B) the location of existing significant cold-water refuges where cooler tributaries enter the mainstem. Additionally, showing the location of CWRs added to low density CWR reaches in the added CWRs thermalscape simulation.

The migration corridor model was constructed in the HexSim development environment (Schumaker and Brookes 2018). Modeled individuals were attributed traits that can change as a result of their interactions with the environment and other organisms. The model was designed to be as mechanistic as possible, given limitations in available empirical data and understanding of the likelihood of a fish’s behavior under various environmental conditions. The model used probabilistic behavioral simulations to simulate conditions where knowledge of primary mechanisms was limited.

The model was used to simulate the influence of behavioral thermoregulation on the energetic status and behavior of the adult upstream migration component of the salmonid life cycle (Figure 1). Additional details are provided in Snyder et al. (2019). Species- and population-specific migration start date and initial body masses were attributed to modeled fish based on empirical data (Jepson et al. 2010; Keefer et al. 2009). Depending on their thermal exposure and tendency for behavioral thermoregulation (Goniea et al. 2006; Keefer et al. 2009) modeled fish moved between the river mainstem and available CWRs within the simulated migration corridor. Refuge location (relative to an individual fish) and the volume of CWR available also influenced refuge use because the model mechanisms incorporated CWR density and distance into the behavioral decision making process. Modeled fish were assigned species-specific targeted arrival times on spawning grounds based on spawn-timing records (Groves and Chandler 1999; Keefer et al. 2018). Individuals’ motivation to use cold-water refuges decreased as the targeted arrival time on spawning grounds neared because the imperative to spawn became the dominant behavioral motivation. Simulated metabolic processes consumed energy at a rate associated with fish mass and thermal exposure based on the Wisconsin Bioenergetics framework (Deslauriers et al. 2017; Plumb 2018; Stewart and Ibarra 1991). The model estimates mass and energy density hourly based solely on energy consumed through respiration since foraging is disrupted during upstream migration. Energy expenditure rates declined during CWR residency, because basal metabolic rates are lower in near-optimum temperatures and also as a consequence of reduced activity costs (Deslauriers et al. 2017; Plumb 2018; Stewart and Ibarra 1991). Bioenergetics model shortcomings stemming from data limitations include a lack of size- and temperature- related activity coefficients parameters and activity costs that correspond directly to water velocity and metabolic power demands. The bioenergetics model also did not account for differences in habitat quality (e.g. dissolved oxygen levels) beyond thermal heterogeneity, nor does it incorporate dynamic hydrologic factors such as discharge rate.

Much of the Columbia River migration corridor has been modified by a series of hydropower dams and their affiliated reservoirs. Although episodic thermal stratification is observed near John Day and McNary dams (US Army Corps of Engineers, unpublished data), reservoir reaches are generally well-mixed vertically (Yearsley 2009). When thermal stratification does occur, the cooler water found at greater depths does not appear to be used by migrating salmonids (Keefer et al. 2019). Low diel variability (<0.5 °C on average) suggests few opportunities exist for nighttime escape from warm temperatures (University of Washington, 2019). In the 288-rkm section of the lower- and mid-Columbia River from Bonneville dam (235 rkm from the Pacific Ocean) to the Snake River confluence (537 rkm) the primary CWRs are formed at confluences where cooler tributaries flow into the Columbia River (Keefer et al. 2018; Snyder et al. 2019).

For our models, we included CWRs that met accessibility and temperature criteria defined by the US Environmental Protection Agency (2020). These criteria included a minimum tributary mean August discharge of > 0.28 m3/s (~10 cfs), corresponding to CWR dimensions (tributary and plume combined) deemed likely to provide suitable depths and volumes for adult salmon and steelhead thermoregulation based upon field surveys. Mean August discharge of tributaries forming CWRs were estimated using USGS gage data, USGS StreamStats, or from NHDPlus depending on availability (US Geological Survey 2016, US Geological Survey 2018). Volumes of CWRs were estimated via a combination of field surveys (depth and temperature) and plume/tributary models that incorporated tributary discharge and confluence morphology (see US Environmental Protection Agency (2020) for additional details). In the simulations, CWR volume estimates were constant through time. A temperature criterion of 2°C less than that of the Columbia River at some point during the migration was applied for consistency and applicability to regional water temperature guidance. These criteria captured seven thermal refuges known to be used by salmon and steelhead based upon prior research (Keefer et al. 2008; Keefer et al. 2018, etc.) and identified two additional locations (Rock Creek, Eagle Creek) for a total of nine modeled CWRs within the stretch of the river examined in this study (US Environmental Protection Agency 2020; Figure 2). In the simulations, CWR availability was assessed hourly based on the magnitude of the difference in temperature between the mainstem Columbia River and the CWR.

We modified the fish populations described in Snyder et al. (2019) in order to represent the life histories of Grande Ronde River summer steelhead (Oncorhynchus mykiss) and Snake River fall Chinook Salmon (Oncorhynchus tshawytscha). Because their migration timing corresponds with warmer Columbia River temperatures, these specific populations have a higher probability of moving to cold-water refuges for behavioral thermoregulation (Jepson et al. 2010; Keefer et al. 2009). While both species have been shown to slow or stop migration or seek thermal refuge as temperatures approach 20–21°C (summarized by Richter and Kolmes 2004), we used observed CWR use patterns from Columbia River migrants (Goniea et al. 2006; Keefer et al. 2009). Generally, fall Chinook Salmon migrate through the Columbia River from August to October and arrive on spawning grounds and spawn between October and December (Groves and Chandler 1999; Jepson et al. 2010). Summer steelhead migrate upstream from June to October, have a protracted migration that includes overwintering, and spring spawning (Keefer et al. 2008; Robards and Quinn 2002). However, most pass through the Columbia River migration corridor by November. Thus, summer steelhead, particularly earlier migrants, are more likely to experience longer periods of higher temperatures during July and August than fall Chinook Salmon and later-migrating summer steelhead that experience less-severe conditions as temperatures tend to cool through September and beyond. For Grande Ronde River summer steelhead, we set the population entry weight and timing to 5092 ±1674 g SE and Aug. 15 ±15 days SE respectively (Keefer et al. 2020; Keefer et al. 2009). For the Snake River fall Chinook Salmon population, we set the entry weight and timing to 4279 ± 2088 g SE and Sept. 3 ± 6.5 SE days respectively (Keefer et al. 2020; Jepson et al. 2010). Entry timing values were based on entry into the simulated migration corridor at the Bonneville Dam forebay. The initial body mass distributions were calculated from the fork lengths of radio tagged fish of known origin, that were converted first to total length and subsequently to mass using published equations (Conrad and Gutmann 1996; Mahmoudi et al. 2014). Other species-specific attributes (e.g., swim speed, propensity for behavioral thermoregulation, spawn timing) were unchanged from the original parameterization in Snyder et al. (2019).

Adult salmonids in the Columbia River migration corridor can spend a large portion of their time swimming through fishways at hydropower structures (Connor et al. 2019; Crozier et al. 2017; Keefer et al. 2004). From the Bonneville Dam forebay to the Snake River confluence, fish navigate past three hydropower dams and four reservoirs. Hydropower passage time was recorded in the simulation as the aggregate of time spent in the tailrace and fishways at each hydropower dam. For migrating individuals, passing through dam tailraces and fishways decreases swim speed and increases energy consumed (Brown et al. 2006). Higher activity costs at hydropower dams were incorporated into the bioenergetics framework by multiplying the energy consumed per hour by scaling factors designed for tailraces (1.62) and fishways (1.26) based on estimated residence times by radio-tagged fish described by Keefer et al. (2017). Arrival time (day or night) and overall passage speed (fast, slow, very slow) were used to set simulated individual’s tailrace behavior (e.g. Crozier et al. 2017). Data from Crozier et al. (2017) were used to develop distributions, characterized by a mean and standard deviation, making it possible to parameterize each of the six arrival timing-passage speed combinations.

Because of the expanded spatial extent and incorporation of hydropower structures, the migration corridor simulation model was re-calibrated after Snyder et al. (2019). Rather than specific populations or years, the model tuning was realized for average recent river temperatures (1992–2016) and for the aggregate Columbia River ‘upriver bright’ fall Chinook Salmon and summer steelhead populations. The model was tuned generally using the same methods as in Snyder et al. (2019), which entailed modification of tuning parameters based on their effect on emergent model outcomes and comparing those to empirical observations. For more details see Appendix A. We also tuned the passage time through modeled tailraces and fishways based on empirical distributions. Fishway passage timing was adjusted so that overall hydropower passage times (tailrace + fishway) emulated the observed median hydropower dam passage times of radio-tagged adult salmon and steelhead (19h) (± 15%) at the Columbia River dams (Keefer et al. 2017).

Experimental design

We developed four experiments to further our understanding of how CWR availability and Columbia River corridor temperatures influence fish condition outcomes during upstream migration. To this end we used our spatial IBM to evaluate alternative future scenarios depicting contrasting refuge configurations and temperature regimes. Our current conditions scenario represented present-day CWR configurations and river temperatures, two additional scenarios examined the significance of altered CWR availability, and a final scenario gauged model response to a cooler Columbia River. We use the labels lost CWRs, added CWRs, and cooler Columbia, to refer to our three alternative future scenarios. Current conditions of the Columbia River were simulated based on water temperature observations at hydropower structures from the year 2017 (University of Washington, 2019). For each reservoir, we modeled temperature as spatially homogeneous except for within the CWRs, which were comprised of tributary and plume components. We obtained current CWR temperatures from averaged available temperature observations made over the past 20 years (Snyder et al. 2019).

We created lost CWRs using the current conditions thermalscape modified so that CWRs were no longer available for behavioral thermoregulation. To examine how configuration of CWRs within the migration corridor influences fish condition, we created added CWRs where we added ‘experimental’ CWRs in reaches where CWRs were the farthest apart or currently absent (hereafter, added CWRs; Figure 1). Five total additional CWRs were added between John Day and McNary Pools, a reach with few existing CWRs. These reservoirs represent a lengthy portion of the migration corridor that has been previously recognized as an area susceptible to high water temperature, potential adult migration blockages, and a distinct lack of available CWRs (Keefer et al. 2018). In the added CWRs scenario, each hypothetical CWR was assigned a volume of 6000 m3 (about twice the volume of the Eagle Creek CWR) and a thermal time series equal to the coldest of the actual refuges present in our system (Little White Salmon River CWR; mean Jul. – Nov. temperature ~9.5 °C).

We designed the cooler Columbia scenario to examine the relative influence of mainstem temperature versus CWR configuration in determining fish passage outcomes. The cooler Columbia thermalscape decreased the present-day mainstem temperature time series by 1°C. This scenario assumed uniform reductions in both day and night time temperatures, and did not account for potential spatial variability in warming (e.g. Steel et al. 2019).

Lowering mainstem Columbia River temperatures is technically possible as is the creation of additional CWRs via restoration of flow or through engineering. However, our simulated scenarios were not constructed with the plausibility of either of those management actions in mind. Rather the overarching objective was to further the understanding of the relative magnitude of effects on fish condition from potential management actions. The lost CWRs was a simulation experiment to illustrate the value of the CWRs to fish condition under the current thermalscape when compared to the current conditions and is not meant to imply a realistic scenario in which CWRs would become entirely unavailable. The added CWRs and cooler Columbia scenarios were designed to illustrate the relative effects of potential management actions of either lowering mainstem Columbia River temperatures or increasing available CWRs.

Fish condition outcomes

Fish condition at the simulated migration terminus (Snake River – Columbia River confluence for this model application) was assessed using a suite of emergent model outcomes capturing both direct physiological exposure to stressful temperatures, and indirect effects of exposure on passage timing and energy consumed. Direct physiological exposure was measured as cumulative degree days where exposure hours greater than 20, 21, and 22°C thresholds wee summed and divided by 24. Indirect exposure outcomes include each fish’s percent of original energy remaining, their migration corridor exit date, and their migration passage time. These fish condition outcomes were intended to link CWR availability and mainstem temperature to survivorship and spawning success. Fish condition outcomes in the simulation model were measured separately. However, we caution that one could expect potential interactions between fish energy content, behavior, heat stress, influence of photo period on migration strategy and other unmeasured stressors on cellular and physiological response to ultimately determine fish condition and eventual spawning success (e.g. Connor et al. 2019; Corey et al. 2017).

Results

Behavioral Outcomes

Species-specific differences in arrival timing and propensity for behavioral thermoregulation led to inter-species differences in migration duration. Snake River fall Chinook Salmon spent less time in the migration corridor than Grande Ronde River summer steelhead with mean passage times under current conditions of 12.4 and 37.5 d, respectively (Table 1). Behavioral thermoregulation increased total migration time for summer steelhead and fall Chinook Salmon between Bonneville Dam and the Snake River confluence. Comparison of migration duration between the current condition and lost CWRs scenarios revealed that behavioral thermoregulation contributed to a higher proportion of migration time in steelhead (mean time in CWR = 21.1 d, or 56% of total migration time) compared to Chinook Salmon (mean = 0.8 d, or 6%). Some individuals remained for long periods in CWRs, as reflected in maximum migration durations of 112 and 62.6 days for steelhead trout and Chinook Salmon respectively. Increasing the density of available cold-water refuges (added CWRs thermalscape) did not change the migration time substantially when compared to the current conditions thermalscape. The cooler Columbia thermalscape decreased mean migration time by 4 d for summer steelhead and by 0.4 d for Chinook Salmon.

Table 1.

Simulated steelhead trout and Chinook Salmon mean residence times (d) for total migration corridor and by corridor component (pools, hydropower structures, cold-water refuges) for each thermalscape.

Grande Ronde River Summer Steelhead
Corridor Pools Hydropower Structures Cold-water Refuges Total Passage
Thermalscape mean range mean range mean range mean range
Current Conditions 13.3 8.9–21.8 3.2 0.5–9.75 21.0 0.0–94.8 37.5 10.1–112.1
Lost CWRs 13.2 9.2–21.0 3.2 0.5–10 . . 16.4 10.3–29.4
Cooler Columbia 13.3 9.1–22.2 3.2 0.5–9.8 17.0 0.0–84.1 33.5 10.2–103.6
Added CWRs 11.7 7.9–19.7 3.2 0.5–9.8 21.7 0.0–94.5 36.6 9.3–113.0
Snake River Fall Chinook Salmon
Corridor Pools Hydropower Structures Cold-water Refuges Total Passage
Thermalscape mean range mean range mean range mean range
Current Conditions 8.4 5.2–14.1 3.2 0.5–9.3 0.9 0.0–48.3 12.4 6.1–62.6
Lost CWRs 8.4 5.3–13.2 3.2 0.4–9.4 . . 11.6 6.1–21.5
Cooler Columbia 8.4 5.2–14.0 3.2 0.5–9.9 0.5 0.0–38.0 12.1 6.3–52.5
Added CWRs 8.4 5.1–14.2 3.2 0.5–10.0 1.6 0.0–41.0 13.2 6.2–56.3
Open in a new tab

Date of arrival at the Snake River confluence is an emergent model outcome resulting from individual propensity for behavioral thermoregulation, availability of CWRs, temperature exposure, swim speed, and hydropower passage timing. Loss of CWRs effectively eliminated migration strategies that use these habitats, and this was reflected in the decreased range of migration corridor exit dates for simulated fall Chinook Salmon and summer steelhead (Figure 3). When CWRs were unavailable, summer steelhead migrated more rapidly and exhibited earlier median exit date from mid-September (current conditions) to late August (lost CWRs).

Figure 3.

Figure 3.

Open in a new tab

Histogram of exit date range across modeled thermalscapes.

The highest density of CWRs under current conditions is found in the Bonneville reservoir where seven of nine CWRs are located (Figure 2). For Chinook Salmon and summer steelhead, CWR residence time in the added CWRs scenario substantially increased in the McNary and John Day reservoirs compared to current conditions (Figure 4). The populations also demonstrated a decrease in CWR residency in the John Day reservoir in the cooler Columbia simulation because the single CWR in that pool, the Umatilla River CWR, was not substantially cooler than the mainstem Columbia for much of the summer (Δ hourly temperature < 2°C) and thus was less frequently identified and used as a potential CWR by the modeled fish.

Figure 4.

Figure 4.

Open in a new tab

For simulated (A) Snake River fall Chinook Salmon and (B) Grande Ronde River summer steelhead mean and confidence intervals (95%) of total population cold-water refuge (CWR) residence time in four reservoirs of the Columbia River for three thermalscapes: current thermalscape with available CWR (current conditions), current thermalscape with added experimental CWRs (added CWRs), and cooler Columbia River (cooler Columbia) thermalscape. Incidental CWR use in the fourth modeled scenario (lost CWRs) not included.

Accumulated thermal exposure

The assessed exposure (cumulative degree days) to warm but sublethal temperatures across a range of indicative thermal tolerance thresholds of 20, 21, and 22°C demonstrated large differences among some scenarios (Figure 5). Comparing the current condition scenario with the lost CWRs demonstrates that the loss of CWRs increases the proportion of the population with higher accumulated degree days for steelhead. For example, at the median (50th percentile) for steelhead under current conditions, CWRs reduce exposure to temperatures >21°C by 120 degree days on average compared to the lost CWRs thermalscape. Maximum cumulative degree days across all thresholds were greater for steelhead than Chinook Salmon. Compared to current conditions, the cooler Columbia scenario was the most effective at decreasing the number of individuals with high accumulated degree days for both species. The effect of cooler Columbia increased as the cumulative degree day threshold increased from 20 to 22°C. For Chinook Salmon, the lost CWRs scenario did not have a large impact on the cumulative degree days. For both species, the added CWRs thermalscape changed accumulated thermal exposure little when evaluated against results from the current conditions and cooler Columbia thermalscapes.

Figure 5.

Figure 5.

Open in a new tab

For each modeled thermalscape, a reverse cumulative distribution function (1-CDF), showing the proportion of individuals exceeding cumulative degree day values (x-axis) given degree day thresholds of 20, 21, and 22 °C (panels A-B, C-D, and E-F, respectively). Figure A,C, and E are derived from modeled Grande Ronde River summer steelhead and B,D, and F, Snake River Fall Chinook Salmon temperature exposure time series. Note that x-axis scales vary between species. Cumulative degree days is calculated as the sum of the hourly temperatures from a modeled individual’s temperature exposure time series above the thresholds (e.g. 20 °C) converted to days by dividing by 24.

Energy Use

Overall, fall Chinook Salmon had higher energy remaining after migration to the Snake River confluence than steelhead. For example, in the current conditions scenario, the estimated remaining somatic energy for fall Chinook Salmon was 80.8% on median, versus 71.6% for summer steelhead (Figure 6). The cooler Columbia scenario resulted in lower migratory energetic costs and higher somatic energy remaining for summer steelhead (median = 82.2%) and fall Chinook Salmon (74.6%) when compared to current conditions with CWRs. Addition or removal of CWRs from the modeled river corridor resulted in negligible effects on net energetic demands for migrating Chinook Salmon and steelhead.

Figure 6.

Figure 6.

Open in a new tab

Boxplot of percent energy remaining for modeled (A) Grande Ronde River summer-run steelhead and (B) Snake River fall-run Chinook Salmon populations through the modeled thermalscapes.

Discussion

The modeling approach used in this study provides a means to evaluate the sufficiency of CWRs to mitigate exposure of upstream migrating Chinook Salmon and steelhead trout in the Columbia River. The interplay between diverse migratory behaviors and spatiotemporal variability in temperatures is well-suited to evaluation using a spatially explicit, individual-based model. Our application of this model to the question of CWR sufficiency complements, and is empirically reliant on, prior research that has documented CWR use by these species (Connor et al. 2019; Goniea et al. 2006; Keefer et al. 2018; Plumb 2018). Scenarios of CWR loss, CWR addition, and cooler migration corridor conditions that we evaluated suggest that accumulated thermal exposure is sensitive to CWR availability (summer steelhead) and Columbia River temperatures (both species), but is influenced less so by the addition of CWRs. In contrast, scenarios we evaluated suggest that net energy remaining in migrants was primarily sensitive to the temperature suitability of the migration corridor.

Previous research has shown that CWRs can provide opportunity for significant relief from exposure to warm river temperatures (e.g. Plumb 2018). In our model scenarios for summer steelhead, loss of opportunity for behavioral thermoregulation in the lost CWRs scenario increased accumulated thermal exposure above all threshold temperatures (Figure 5 A,C,E). Therefore, behavioral thermoregulation was an effective strategy for steelhead to decrease accumulated thermal exposure under the current configuration of CWRs. For summer steelhead, CWR use was effective at reducing thermal exposures for all the temperature thresholds. This suggests that CWRs along the Columbia River migration corridor could be increasingly important under warmer conditions. In contrast, fall Chinook Salmon demonstrated little difference in accumulated thermal exposure between the current conditions and loss of CWRs thermalscapes (Figure 5 B,D,F), which suggests behavioral thermoregulation was not a particularly effective strategy for decreasing chronic exposure given the current availability of CWRs.

Results of the added CWRs and lost CWRs experiments suggest that accumulated thermal exposure was highly sensitive to the loss of existing CWR volume available but was less sensitive to additional refuges along the migration corridor. For summer steelhead and fall Chinook Salmon, the thermalscape with cooler Columbia River temperature was the most effective at reducing the proportion of the population with higher accumulated degree days. Addition of CWRs increased opportunities for summer steelhead to avoid thermally stressful conditions in the John Day and McNary reservoirs. However, the effect on overall exposure to high temperatures was small compared to the currently available CWRs and the effect of lowering Columbia River temperatures (Figure 5). Decreasing the Columbia River temperatures could facilitate reduced chronic exposure to warm river temperatures for summer steelhead and fall Chinook Salmon populations although management options are currently limited. Additional opportunities for CWR use (CWR addition, e.g. Kurylyk et al. 2015) may have the potential to decrease exposure to stressful river temperatures for summer steelhead populations but is unlikely to reduce exposure for fall Chinook Salmon populations.

The ability to maintain adequate stores of energy to fuel migration and spawning is an important variable influencing reproductive condition in migratory salmonids (Robards and Quinn 2002). Depletion of lipids to values below 4 kJ g−1 are associated with mortality and failed spawning success, and thus are a concern (Penney and Moffitt 2014; Plumb 2018). In our scenarios, which ended at the Snake River confluence with the Columbia River and thus captured only part of the full migration route, net energy loss in migrants was sensitive to the temperature suitability of the migration corridor. Regardless of CWR availability, fish must navigate the full length of the corridor, and cannot avoid exposure to the energetic demands associated with passage at each hydropower facility. Thus, mainstem corridor conditions set a base level of net energetic demand shared by all fish, regardless of CWR availability or use. Our modeled Chinook Salmon and steelhead spent on average 67 and 36% of their total travel time in the corridor reservoirs, 25 and 8% of their time navigating the corridor hydropower facilities (fishways and tailraces), and the remaining 8 and 56% of their time in CWRs, respectively. Simulated temperature reductions of the Columbia River corridor resulted in slight increases in net energy remaining for both steelhead trout and fall Chinook Salmon (Figure 6) which reflected the predominant influence of the corridor conditions on net energy expenditure. As a result, it appears that opportunities to reduce energetic costs of migrating adult salmon and steelhead lies primarily in the Columbia River and Snake River corridors, either via altered temperatures (Keefer et al. 2018; Plumb 2018), reduced energetic costs of navigating hydropower structures (Brown et al. 2006; Caudill et al. 2007; Keefer and Caudill 2016), or by the fish themselves, perhaps, by altering migration timing to reduce exposure to warm corridor temperatures (Crozier et al. 2008, Goniea et al. 2006, Keefer et al. 2008).

The inter-specific differences in outcomes observed between Chinook Salmon and steelhead trout reflect differences in each species’ life histories and associated propensity for CWR use. In previous studies, radio-tagged Chinook Salmon were observed to use CWRs at tributary mouths less frequently, and for shorter durations than steelhead trout (Goniea et al. 2006; Keefer et al. 2018), and this was incorporated into our parameterization (Snyder et al. 2019). Because of this, effects of changing CWR availability on exposure to high temperatures differed between Chinook Salmon and steelhead trout (Figure 5). Steelhead trout responded to both additions and loss of CWRs, whereas Chinook Salmon responded primarily to Columbia River corridor temperatures, particularly at higher temperatures (Figures 5D, 5F). Loss of CWRs likely had little effect on Chinook Salmon thermal exposures due to their relatively short duration use of these habitats. Loss of CWRs had a substantial effect on steelhead trout migration duration (Table 1) and thermal exposure (Figure 5). Loss of CWRs resulted in shorter overall migration duration, and thus increased net somatic energy remaining, but increased exposures to acute thermal stress, for steelhead. Although these results are from modeled individuals, they point to potentially strong selective pressures that can act to influence migration timing within and between these two species (Brannon et al. 2004; Waples and Lindley 2018).

Without CWRs, simulated steelhead trout must continue upstream along the migration corridor to seek cold-water. Earlier arrival at the Snake River could be particularly problematic for Grande Ronde River steelhead and other steelhead populations that migrate mid-summer (e.g. Keefer et al. 2009) given warm water temperatures in the Snake and lower Grande Ronde rivers during late August and September. Use of CWRs allows these steelhead to delay arrival to the Snake River, reducing subsequent exposures to potentially stressful temperatures (Keefer and Caudill 2016). Loss of CWRs along the migratory corridor could represent a significant constraint on populations faced with lengthy, and thermally challenging portions of their migrations awaiting them upstream of our end-of-model reach.

Limitations

Application of these models reveals the complex interplay of migration timing, CWR use, and a spatially and temporally dynamic thermal environment on thermal exposure along the migratory pathways used by the study populations. Energy remaining, accumulated thermal exposure, and migration timing outcomes illustrate that while CWR use is clearly important, it cannot be assumed to be universally beneficial. While the HexSim modeling approach was unable to fully capture all integrated effects of thermal exposure, disease exposure, predator encounters, migratory timing, behavioral plasticity, and other delayed effects, the results clearly show that CWRs can in some cases provide critical relief from thermally stressful conditions and allow a greater diversity of migration tactics and migration timing among Columbia River salmon and steelhead populations.

While our modeling approach provides a starting point for assessing sufficiency of CWRs along a migratory corridor, fully disentangling trade-offs will require a better understanding of the ultimate condition consequences of accumulated sub-lethal thermal exposures and associated variations in migration behaviors and arrival timing at the spawning grounds. For example, simulated fish migration behavior might not capture the full variability of behaviors because of limited CWR observations at temperatures greater than 22°C. Because we focus here on the relative fitness-related consequences of broad categories of future riverscape change, we are able to emphasize the strengths inherent in forecasting models such as ours. While using this class of tools to make precise quantitative projections necessitates a high degree of confidence in critical input parameters (with criticality being assessed via sensitivity analysis), well-designed forecasting models can be quite good at evaluating the nature of management trade-offs such as those illustrated in this study (e.g. McLane et al. 2011). Our work grew out of a need for an objective and defensible interpretation of the concept of suitability, in the context of CWRs and the upstream migration of highly impacted salmonid species. As is often the case, we did not have the luxury of waiting years or decades for better data sets to become available, nor were necessary resources available for conducting extensive sensitivity analyses on a multi-factor behavioral model. Our approach addresses the management questions at hand, while also illustrating the novel technology and methodology we hope to share with other researchers conducting similar studies. Promising avenues of additional research include incorporation of physiological tags to better track and evaluate physiological, energetic, and behavioral status of migrants (e.g. Cooke et al. 2008) and associations with CWR use. The energetic demands of navigating fishways and tailraces through the hydropower system comprise a substantial portion of the total energy budget for some migratory stocks, and understanding the energetic demands could benefit from more refined estimates of temperature-dependent passage costs (Caudill et al. 2007; Connor et al. 2019). Additional research is also warranted for better understanding of the physical processes associated with CWR formation and maintenance. Improved models for predicting CWR volume and temporal variability would enhance the sensitivity of our fish modeling results to hydrologic or energy balance changes associated with climate change or river management.

Conclusion

Resource management agencies tasked with managing water quality and fish populations seek to know whether existing CWRs in the Columbia River corridor are ‘sufficient’ to protect migratory salmon and steelhead from adversely warm water temperatures during migration (US Environmental Protection Agency 2020). Such determination is needed in order to evaluate the effectiveness of narrative criteria such as Oregon’s CWR provision (Box 1). In the example illustrated in this paper, present-day CWRs primarily located within the Bonneville reservoir portion of the Columbia River migratory corridor provide opportunities for Chinook Salmon and steelhead trout to reduce exposure to temperatures exceeding 20°C. This effect was slightly enhanced when CWRs were experimentally added to portions of the corridor currently lacking CWRs but not nearly as substantial as decreasing the temperature of the Columbia River. These reductions in exposure could translate to significant improvements in gamete viability and adult pre-spawn survival (Bowerman et al. 2017a; Fenkes et al. 2016).

Water quality standards that attempt to incorporate CWRs into narrative criteria can be informed by approaches such as this one. Simulation modeling approaches allow for in silico experiments that track individual exposure through time and space that would otherwise not be possible. This modeling approach could be used to further explore how starting condition of fish, changing phenology, or propensity for behavioral thermoregulation would influence fish conditions outcomes. Such experiments could also modify the temperature exposure of the migration corridor or the CWRs to simulate future climatic changes (i.e., river warming) or restoration of CWRs to further understanding of temperature exposure on fish condition. Applications will likely require a case-by-case approach, since, as this study illustrates, sufficiency is not only a function of CWR temperature and volume, but is also contingent upon CWR spatial arrangement and is strongly mediated by conditions in the migratory corridor.

Supplementary Material

Supplement1

Acknowledgements

The information in this document was funded in part by the U.S. Environmental Protection Agency. It has been subjected to review by the Center for Public Health and Environmental Assessment’s Pacific Ecological Systems Division and approved for publication. Approval does not signify that the contents reflect the views of the Environmental Protection Agency, nor does mention of trade names or commercial products constitute endorsement or recommendation for use by the U.S. Government. The information in this document has been approved by, and does represent the views of the USGS. This paper has been peer reviewed and approved for publication consistent with USGS Fundamental Science Practices (http://pubs.usgs.gov/circ/1367/). We would also like to acknowledge Antóin M. O’Sullivan and an anonymous reviewer who have commented on previous versions and helped to improve it.

Literature cited

  1. Bernatchez L, and Dodson JJ. 1987. Relationship between bioenergetics and behavior in anadromous fish migrations. Canadian Journal of Fisheries and Aquatic Sciences 44(2):399–407. [Google Scholar]
  2. Bowerman T, Pinson‐Dumm A, Peery C, and Caudill C. 2017a. Reproductive energy expenditure and changes in body morphology for a population of Chinook salmon Oncorhynchus tshawytscha with a long distance migration. Journal of Fish Biology 90(5):1960–1979. [DOI] [PubMed] [Google Scholar]
  3. Bowerman T, Roumasset A, Keefer ML, Sharpe CS, and Caudill CC. 2017b. Prespawn mortality of female Chinook Salmon increases with water temperature and percent hatchery origin. Transactions of the American Fisheries Society 147(1):31–42. [Google Scholar]
  4. Brannon EL, Powell MS, Quinn TP, Talbot A. 2004. Population structure of Columbia River Basin Chinook salmon and steelhead trout. Reviews in Fisheries Science 12(2–3):99–232. [Google Scholar]
  5. Brown RS, Geist DR, and Mesa MG. 2006. Use of electromyogram telemetry to assess swimming activity of adult spring Chinook salmon migrating past a Columbia River dam. Transactions of the American Fisheries Society 135(2):281–287. [Google Scholar]
  6. Caudill CC, and coauthors. 2007. Slow dam passage in adult Columbia River salmonids associated with unsuccessful migration: delayed negative effects of passage obstacles or condition-dependent mortality? Canadian Journal of Fisheries and Aquatic Sciences 64(7):979–995. [Google Scholar]
  7. Connor WP, and coauthors. 2019. Upstream migration and spawning success of Chinook salmon in a highly developed, seasonally warm river system. Reviews in Fisheries Science and Aquaculture 27(1):1–50. [Google Scholar]
  8. Conrad RH, and Gutmann JL. 1996. Conversion equations between fork length and total length for chinook salmon (Oncorhynchus tshawytscha). Northwest Indian Fisheries Commission. [Google Scholar]
  9. Cooke SJ, and coauthors. 2008. Developing a mechanistic understanding of fish migrations by linking telemetry with physiology, behavior, genomics and experimental biology: an interdisciplinary case study on adult Fraser River sockeye salmon. Fisheries 33(7):321–339. [Google Scholar]
  10. Corey E, Linnansaari T, Cunjak RA, and Currie S. 2017. Physiological effects of environmentally relevant, multi-day thermal stress on wild juvenile Atlantic salmon (Salmo salar). Conservation Physiology 5(1). [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Crossin GT, and coauthors. 2008. Exposure to high temperature influences the behaviour, physiology, and survival of sockeye salmon during spawning migration. Canadian Journal of Zoology 86(2):127–140. [Google Scholar]
  12. Crozier L, and coauthors. 2008. Potential responses to climate change in organisms with complex life histories: evolution and plasticity in Pacific salmon. Evolutionary Applications 1(2):252–270. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Crozier LG, Bowerman TE, Burke BJ, Keefer ML, and Caudill CC. 2017. High‐stakes steeplechase: a behavior‐based model to predict individual travel times through diverse migration segments. Ecosphere 8(10). [Google Scholar]
  14. Crozier LG, and coauthors. 2019. Climate vulnerability assessment for Pacific salmon and steelhead in the California Current Large Marine Ecosystem. PLoS One 14(7). [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Deslauriers D, Chipps SR, Breck JE, Rice JA, and Madenjian CPJF. 2017. Fish bioenergetics 4.0: an R-based modeling application. 42(11):586–596. [Google Scholar]
  16. Fenkes M, Shiels HA, Fitzpatrick JL, and Nudds RL. 2016. The potential impacts of migratory difficulty, including warmer waters and altered flow conditions, on the reproductive success of salmonid fishes. Comparative Biochemistry and Physiology Part A: Molecular & Integrative Physiology 193:11–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Frechette DM, Dugdale SJ, Dodson JJ, and Bergeron NE. 2018. Understanding summertime thermal refuge use by adult Atlantic salmon using remote sensing, river temperature monitoring, and acoustic telemetry. Canadian Journal of Fisheries and Aquatic Sciences 75(11):1999–2010. [Google Scholar]
  18. Fullerton AH, and coauthors. 2017. Longitudinal thermal heterogeneity in rivers and refugia for coldwater species: effects of scale and climate change. Aquatic Sciences 80(1):3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Geist DR, and coauthors. 2006. Survival, development, and growth of fall Chinook salmon embryos, alevins, and fry exposed to variable thermal and dissolved oxygen regimes. Transactions of the American Fisheries Society 135(6):1462–1477. [Google Scholar]
  20. Goniea TM, and coauthors. 2006. Behavioral thermoregulation and slowed migration by adult fall Chinook salmon in response to high Columbia River water temperatures. Transactions of the American Fisheries Society 135(2):408–419. [Google Scholar]
  21. Gross MR, Coleman RM, McDowall RM 1988. Aquatic Productivity and the Evolution of Diadromous Fish Migration. Science 239(1291–1293). [DOI] [PubMed] [Google Scholar]
  22. Groves PA, and Chandler JA. 1999. Spawning habitat used by fall Chinook salmon in the Snake River. North American Journal of Fisheries Management 19(4): 912–922. [Google Scholar]
  23. Jepson M, Keefer M, Naughton G, Peery C, and Burke BJ. 2010. Population composition, migration timing, and harvest of Columbia River Chinook salmon in late summer and fall. North American Journal of Fisheries Management 30(1):72–88. [Google Scholar]
  24. Kaushal SS, and coauthors. 2010. Rising stream and river temperatures in the United States. Frontiers in Ecology and the Environment 8(9):461–466. [Google Scholar]
  25. Keefer ML, and Caudill CC. 2016. Estimating thermal exposure of adult summer steelhead and fall Chinook salmon migrating in a warm impounded river. Ecology of Freshwater Fish 25(4):599–611. [Google Scholar]
  26. Keefer ML, Clabough TS, Jepson MA, Bowerman T, and Caudill CC. 2019. Temperature and depth profiles of Chinook salmon and the energetic costs of their long-distance homing migrations. Journal of Thermal Biology 79:155–165. [DOI] [PubMed] [Google Scholar]
  27. Keefer ML, and coauthors. 2018. Thermal exposure of adult Chinook salmon and steelhead: Diverse behavioral strategies in a large and warming river system. PloS one 13(9). [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Keefer ML, and coauthors. 2017. Migration of adult salmonids in the federal Columbia River hydrosystem: a summary of radiotelemetry studies, 1996–2014. United States Army Corps of Engineers. Portland District. [Google Scholar]
  29. Keefer ML, Jepson MA, Clabough TS, and Caudill CC. 2020. Migration of adult salmonids in the federal Columbia River hydrosystem: a summary of radiotelemetry studies, 1996–2016. Northwest Knowledge Network Data Repository. DOI: 10.7923/w61y-qs53. [DOI] [Google Scholar]
  30. Keefer ML, Peery C, Jepson M, and Stuehrenberg L. 2004. Upstream migration rates of radio‐tagged adult Chinook salmon in riverine habitats of the Columbia River basin. Journal of Fish Biology 65(4):1126–1141. [Google Scholar]
  31. Keefer ML, Peery CA, and High B. 2009. Behavioral thermoregulation and associated mortality trade-offs in migrating adult steelhead (Oncorhynchus mykiss): variability among sympatric populations. Canadian Journal of Fisheries and Aquatic Sciences 66(10):1734–1747. [Google Scholar]
  32. Keefer ML, Boggs CT, Peery CA, and Caudill CC. 2008. Overwintering distribution, behavior, and survival of adult summer steelhead: variability among Columbia River populations. North American Journal of Fisheries Management 28(1):81–96. [Google Scholar]
  33. Kurylyk BL, MacQuarrie KT, Linnansaari T, Cunjak RA, and Curry RA. 2015. Preserving, augmenting, and creating cold‐water thermal refugia in rivers: Concepts derived from research on the Miramichi River, New Brunswick (Canada). Ecohydrology 8(6):1095–1108. [Google Scholar]
  34. Mahmoudi R, Soltani M, Matinfar A, Gilkolai SR, and Kamali A. 2014. Morphometric relationship between length- weight, length-length and condition factor in farmed rainbow trout (Oncorhynchus mykiss). Bulletin of Environment, Pharmacology and Life Sciences 3(4):215–220. [Google Scholar]
  35. McLane AJ, Semeniuk C, McDermid GJ, and Marceau DJ. 2011. The role of agent-based models in wildlife ecology and management. Ecological Modelling 222(8): 1544–1556. [Google Scholar]
  36. McClure MM, Holmes EE, Sanderson BL, and Jordan CE. 2003. A large‐scale, multispecies status assessment: anadromous salmonids in the Columbia River basin. Ecological Applications 13(4):964–989. [Google Scholar]
  37. Northcote T 1984. Mechanisms of fish migration in rivers. Pages 317–358 in McCleave JD, Arnold GP, Dodson JJ, and Neill WH, editor. Mechanisms of migration in fishes. Plenum, New York. [Google Scholar]
  38. Penney ZL and Moffitt CM. 2014. Proximate composition and energy density of stream-maturing adult steelhead during upstream migration, sexual maturity, and kelt emigration. Transactions of the American Fisheries Society 143(2):399–413. [Google Scholar]
  39. Plumb JM 2018. A bioenergetics evaluation of temperature‐dependent selection for the spawning phenology by Snake River fall Chinook salmon. Ecology and evolution 8(19):9633–9645. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Poole GC, and coauthors. 2004. The Case for Regime-based Water Quality Standards. Bioscience 54(2):155–161. [Google Scholar]
  41. Richter A and Kolmes SA. 2005. Maximum temperature limits for Chinook, coho, and chum salmon, and steelhead trout in the Pacific Northwest. Reviews in Fisheries Science 13(1):23–49. [Google Scholar]
  42. Robards MD, and Quinn TP. 2002. The migratory timing of adult summer-run steelhead in the Columbia River over six decades of environmental change. Transactions of the American Fisheries Society 131(3):523–536. [Google Scholar]
  43. Schumaker NH, and Brookes A. 2018. HexSim: a modeling environment for ecology and conservation. Landscape Ecology 33(2):197–211. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Shaftel R, and coauthors. 2020. Thermal Diversity of Salmon Streams in the Matanuska‐Susitna Basin, Alaska. Journal of the American Water Resources Association. [Google Scholar]
  45. Snyder MN, and coauthors. 2019. Individual based modeling of fish migration in a 2-D river system: model description and case study. Landscape ecology 34(4):737–754. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Steel EA, and coauthors. 2019. Thermal landscapes in a changing climate: biological implications of water temperature patterns in an extreme year. Canadian Journal of Fisheries and Aquatic Sciences 76(10):1740–1756. [Google Scholar]
  47. Stewart DJ, and Ibarra M. 1991. Predation and production by salmonine fishes in Lake Michigan, 1978–88. Canadian Journal of Fisheries and Aquatic Sciences 48(5):909–922. [Google Scholar]
  48. Torgersen CE, Ebersole JL, and Keenan DM. 2012. Primer for identifying cold-water refuges to protect and restore thermal diversity in riverine landscapes. Environmental Protection Agency, EPA 910-C-12–001, Seattle, WA. [Google Scholar]
  49. University of Washington. 2019. Fish Passage Data. Columbia River DART (Data Access Real Time). https://www.cbr.washington.edu/dart
  50. US Environmental Protection Agency. 2003. EPA Region 10 guidance for Pacific Northwest state and tribal temperature water quality standards. EPA 910-B-03–002.
  51. US Environmental Protection Agency. 2020. Columbia River Cold Water Refuges. Available at: https://www.epa.gov/columbiariver/columbia-river-cold-water-refuges. Access May 26, 2020.
  52. US Geological Survey. 2018. National Hydrography Dataset (ver. USGS National Hydrography Dataset Best Resolution (NHD) for Hydrologic Unit (HU), Online at URL https://www.usgs.gov/core-science-systems/ngp/national-hydrography/access-national-hydrography-products, accessed October 1, 2018.
  53. US Geological Survey. 2016. The StreamStats program. Online at http://streamstats.usgs.gov, accessed October 10, 2018.
  54. Waples RS and Lindley ST. 2018. Genomics and conservation units: The genetic basis of adult migration timing in Pacific salmonids. Evolutionary Applications 11(9): 1518–1526. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Webb BW, and Nobilis F. 2007. Long-term changes in river temperature and the influence of climatic and hydrological factors. Hydrological Sciences Journal 52(1):74–85. [Google Scholar]
  56. Yearsley JR 2009. A semi‐Lagrangian water temperature model for advection‐dominated river systems. Water Resources Research 45(12). [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement1
NIHMS1690098-supplement-Supplement1.docx (88.3KB, docx)

RESOURCES